Spent media analysis is an important tool used to provide crucial information about the amino acids and nutrients that cells use during growth phase. Spent media analysis also provides details about the metabolites that are produced as a result of cell growth. This is useful to cell line and process development teams in that it provides data on media composition performance, optimal harvest time and product quality changes that can occur during culture.

Dr. Elsa Gorre, BioTherapeutics Development, Janssen Research and Development, gave a very informative talk on this subject at the ASMS Annual Meeting. The talk, “Analysis of Spent Media Metabolites on the SCIEX QTRAP® 6500+ LC-MS/MS System to Assure Product Quality During Cell Line Development for Biologics,” described in detail Janssen’s implementation of spent media analysis via mass spectrometry as part of their clone selection process. They were able to verify sequence variant misincorporation and to monitor the amino acid and metabolite profiles of different clones. I have summarized the highlights of the talk in this article.

Spent Media Analysis Purpose

Dr. Gorre began the talk by explaining the purpose of her team’s spent media analysis work. They are primarily interested in monitoring daily changes in amino acids and metabolites to optimize growth conditions for clones, promote high cell viability, maximize production of titer, ensure high product quality, and reduce or eliminate sequence variant misincorporations. She said that the Janssen proprietary media development team is continuously modifying the composition of media to minimize depletion of essential amino acids and that they are using a modeling approach to predict amino acid and metabolite trends in real time.

The team is also using spent media analysis for verification of sequence variants. Sequence variants (SVA) are unintended changes in the amino acid’s sequence that could potentially affect efficacy, safety and immunogenicity of the final product. Causes of SVA include DNA mutation either through replication error or DNA damage and mRNA mistranslation events including codon misreading (most common) or tRNA mischarging.

Spent Media Analysis Challenges

Dr. Gorre then described the cell line development group’s sampling process and shared her views as to the challenges of spent media analysis. She cites sample preparation (specifically extraction methods), having to analyze different classes of analytes using only one method (i.e. amino acids vs. vitamins vs. nutrients), and the large range in concentration of the different analytes. To address these challenges the team compared extraction methods using NMR technology. For sample one they applied acetonitrile metabolite methods of extraction and for sample two they applied methanol metabolite extraction. The results showed that metabolites had 10-20% difference in concentration between the two extraction methods.

Scheduled Multiple Reaction Monitoring to Increase MRM Method Efficiency

The team also explored using the Scheduled Multiple Reaction Monitoring (MRM) algorithm from SCIEX. They were aiming to improve MRM method efficiency by maximizing analyte utilization with scheduled MRM.

They found that scheduled MRM improved efficiency in the following ways:

• Each MRM monitored only across its expected elution time
• Decreased concurrent MRMs
• Maintained cycle time and dwell time
• Increased effective duty cycle for every analyte
• Maintained analytical precision

When comparing Normal MRMs vs. Scheduled MRMs, they found:

• Normal MRM resulted in poor peak sampling making robust peak integration very difficult
• Scheduled MRM provided proper cycle times for good peak sampling and increased assay reproducibility from 12% to 7.8% CV

Collaboration with SCIEX for Spent Media Analysis using SCIEX Scheduled MRM Method Workflow

Dr. Gorre then discussed their collaboration with SCIEX to monitor components and metabolites in cell culture and spent media using SCIEX’s targeted scheduled MRM method.

Instrumentation used to analyze media components included:

• Exion LC™ system to achieve high efficient separation
• Phenomenex Kinetex^® F5 column to separate a wide range of compound classes
• QTRAP^® 6500+ LC-MS/MS system for fast scanning and polarity switching – large dynamic range up to 6 fold
• Scheduled MRM™ Algorithm Pro to optimize cycle times and maximize dwell time.

While it is still the beginning of the collaboration, samples for spent media analysis have been sent to SCIEX and separation was performed on the Exion LC™ system with the Phenomenex Kinetex® F5 reverse phase column. They found that they were able to separate the samples into different classes, i.e. metabolites, vitamins, and amino acids. For example, they had experienced difficulty in separating cystine vs. cysteine in spent media analysis, however with the Kinetex F5 reverse phase column they were able to separate the two.

Hydrophilic Interaction Chromatography (HILIC) for Spent Media Analysis

Next Dr. Gorre described another approach to spent media analysis, Hydrophilic Interaction Chromatography (HILIC). They are currently using the Advance BIO MS Spent Media HILIC column that has a zwitterionic phase bonded onto superficially porous silica particles.

She explained that the SCIEX method can measure 111 analytes, but right now they are only monitoring a subset of about 40 analytes. The group that they are collecting data for dictates the number of analytes monitored. Currently they are monitoring the concentration of 40 analytes (metabolites and all 20 amino acids). The modified list is made up of 30 analytes monitored in positive mode and over 10 analytes monitored in negative mode. In combination more than 80 transitions are monitored using the scheduled method. She then shared data on the HILIC chromatography comparison standard vs. sample. In the standard they saw great separation and great response, while in the sample the range in intensity of these peaks was huge. She concluded by saying that the dynamic range of the instrument is very important to their work.

Software for Data Processing

Dr. Gorre said that they are currently using SCIEX OS-Q (SCIEX OS Analytics) software for target analysis and SCIEX MarkerView™ software for global sample comparison by PCA. They will use the software to quantify and compare components during biotherapeutics production. Right now they are looking forward to using these to conduct statistical analysis of the data.

She then presented examples of external calibration curves and said that the system is quite good and that they are able to generate calibration curves for all 40 of the metabolites that they are monitoring.

Specific Analyte Monitoring

One analyte that Dr. Gorre has found is quite important to monitor is ethanolamine. She shared that free ethanolamine is present in cell media and is crucial in cell proliferation. It is the head group of phosphatidylethanolamine (PE) and PE is a major phospholipid in cell membranes. Ethanolamine deficiency results in altered phospholipid composition of the membrane, which affects growth factor signals. It is involved in nutrient storage and transport, as well as signal transduction.

Dr. Gorre and her team followed serine and ethanolamine trends. They found that initially ethanolamine is quite high, but by day 5 ethanolamine drops and there is a deficiency. They found because the decarboxylation of serine produces ethanolamine, that cells are able to “create their own” ethanolamine after depletion at day 5. As a result, by day 7 serine is also depleted.

Another important metabolite to monitor is choline. Methylation of ethanolamine leads to choline. Choline is important for structural integrity of the cell membrane as it is used in synthesis of other phospholipids. Choline is also important in cell signaling as choline containing phospholipids are intracellular messenger molecules. Lastly, choline is a major source of methyl groups. In monitoring choline, they found that it starts high in culture, but is depleted by day 9.

Case Study – Correlating the depletion of amino acids to the sequence variant observed using spent media analysis

This work led Dr. Gorre and her team to look for correlation between the depletion of amino acids and sequence variants (SVA). She began by describing Janssen’s clone selection process. Janssen selects 15 clones to monitor and create a table of different measurements that are used throughout the 16 days of growth in culture. They measure titer (harvest value) and percent viability. Their Mass Spec Group looks at product quality, specifically major sequence variants and glycation percent. Their CDS analytical group looks at fragmentation, percent aggregation, monomer and purity.

The team uses a peptide map to identify sequence variants. They have a threshold for SVA of 0.1%, under which they don’t report. In the case study, they discovered a major SVA – C93Y, where the cystine at position 93 is actually an added tyrosine. They found that several clones showed the same sequence variant, but at different levels. This confirmed that the SVA wasn’t caused by a DNA mutation because when it is a DNA mutation they only see it on one or two clones.

Correlation of sequence variant to amino acids depletion

Sequence variant analysis is acquired on harvest material. They compared the trend plots of clones with the SVA and clones without the SVA. The trend plots show the time point that the depletion of cystine and tyrosine starts. They find that cystine is depleted by day 9 and that tyrosine levels are consistent until they drop at day 9, the same time that the cystine drops in the clone with the SVA. This shows that tyrosine in the media is being depleted and is added to the amino acid sequence of the monoclonal antibody. Then levels of both cystine and tyrosine increase past day 11 due to a cystine and tyrosine feed delivered every other day. The clone without the SVA doesn’t have the decline in cystine or tyrosine.

Conclusion

Dr. Gorre concluded that spent media assay using the SCIEX system allowed for the quantitative measure of different classes of metabolites and amino acids in a 20 minute LC-MS run. It provided a rapid method to analyze over 110 analytes simultaneously and monitor 2 or more transitions per analyte.

They are implementing this spent media analysis as part of their mass spectrometry clone selection package. The most important use for them at this time is for verification of sequence variant misincorporations observed in peptide map analysis. They are also monitoring amino acid and metabolite profiles of different clones and are looking into using statistical methods to look for trends correlate with titer values.